Remote Sensing and GIS
INTRODUCTION
Remote Sensing is the science of making inferences about material objects from measurements, made at a distance, without coming into physical contact with the objects under study. When viewed in this context, any force field - gravity, magnetic, electromagnetic could be used for remote sensing, covering various disciplines from astronomy to laboratory testing of materials. However, currently the term remote sensing is used more com­monly to denote identification of earth features by detecting the characteris­tic electromagnetic radiation that is reflected and or emitted by the earth surface. Every object reflects/scatters  a portion of the electromagnetic energy incident on it depending upon its physical properties. In addition objects emit radiation depending on their temperature and emissivity. If we study the reflectance/emittance of any object at different wavelengths, we get a reflectance/emittance pattern which is characteristic of that object - this is called ‘Spectral signature’. It is like finger prints. Just as we are able to use the finger prints to identify a person, the spectral signatures enable, in principle, to identify the objects.
Visual perception of objects is the best example of remote sensing. We see an object by the light reflected from the object falling on the human eye. Here, eye is the sensor and the nervous system carries information to the brain, which interprets the information in terms of the identification and location of the objects seen. Modern remote sensing is an extension of this natural phenomenon. However, apart from visible light, the electromagnetic radiation extending from the ultraviolet to the far infrared (IR) and the microwave regions are also used for remote sensing of the earth resources. Though the remote sensing techniques were first used operationally for meteorological applications, the present paper emphasises earth resources applications.
A remote sensing system consists of a sensor to collect the radiation and a platform - an aircraft, balloon, rocket, satellite or even a ground-based sen­sor-supporting stand - on which a sensor can be mounted. The information received by the sensor is suitably manipulated and transported back to the earth - may be telemetered as in the case of unmanned spacecraft, or brought back through films, magnetic tapes, etc as in aircraft or manned spacecraft systems. The data are reformatted and processed on the ground to produce either photographs, or computer compatible magnetic tapes (CCT). The photographs/CCTs are interpreted visually/digitally  to produce  thematic maps and other resources information.
SOLAR AND TERRESTRIAL RADIATION
Electromagnetic radiation spans a large spectrum of wavelengths right from very short wavelength gamma rays (10-10m) to long radio waves (106m). In remote sensing, the most useful regions are the visible (0.4 to 0.7 mm), the reflected IR (0.7 to 3 mm), the thermal IR (3 to 5 mm and 8 to 14 mm) and the microwave regions (0.3 to 300 cm). The sun is the important source of electromagnetic radiation used in conventional optical remote sensing. The sun may be assumed to be a blackbody with surface temperature around 60000 K. Maximum of the sun’s radiation occurs around 0.55 mm which is in the visible region. The solar radiation reaching the surface of earth is modi­fied by the atmospheric effects as discussed in the next section. It is ob­served that all bodies at temperatures above zero degrees absolute emit electromagnetic radiation at different wavelengths, as per Planck’s law. The earth can be treated as a blackbody at ~ 3000 K emitting electromagnetic radiation with peak emission at  around 9.7 mm. Figure 1 shows the spectral distribution of reflected solar and self-emitted thermal radiation. According to Planck’s law, the radiation emitted by the earth (3000 K) is much less at all wavelengths compared to that emitted by the sun (60000 K). However, at the earth’s surface because of the great distance between the sun and the earth, the energy in the 7.0 to 15 mm wavelength region is predominantly due to thermal emission of the earth
Atmospheric Effects
In passing through the atmosphere, electromagnetic radiation is scattered and absorbed by gases and particulates. Besides the major atmospheric gaseous components such as molecular Nitrogen and Oxygen, other constitu­ents like Methane, Hydrogen, Helium, and Nitrogen compounds play an im­portant role in modifying the incident radiation energy spectrum. The stron­gest absorption occurs at wavelengths shorter than 0.3 mm primarily due to ozone. There are certain spectral regions where the electromagnetic radiation is passed through without much attenuation and these are called atmospher­ic windows (Figure 2). Remote sensing of the earth’s surface is generally con­fined to these wavelength regions. Atmospheric windows used for remote sensing are 0.4-1.3, 1.5-1.8, 2.2-2.6, 3.0-3.6, 4.2-5.0, 7.0-15.0 mm and 10 mm to 10 cm wavelength regions of electromagnetic spectrum.  Even in  the atmospheric window regions, scattering by the atmospheric constituents produces spatial redistribution of energy. The three important scattering mechanisms are the Rayleigh scattering, Mie scattering and non-selective scattering
A sensor sees the energy reflected from the target and the scattered radia­tion entering its field of view. The radiance measured at the top of the atmosphere has the contributions from (i) single/multiple scattering by the atmospheric constituents  and  reaching  the  field  of  view  (FOV)   of the  sensor (La), (ii) the diffused downward radiation  produced  by  scattering  which  is  reflected  by  the target of interest (Lb),  (iii) the downward component reflect­ed by an adjacent target and further scattered by the atmosphere to get into the FOV (Lc), and reflectance of the target by the direct solar radiation and attenuated to reach the top of the atmosphere - the actual information (LT) (Figure 3). La + Lb + Lc is usually called the path radiance. The path radiance reduces the image contrast (visually the sharpness of the image is reduced). In addition it produces radiometric error, since the information characteristic to target LT is corrupted. Thus the apparent radiance of the ground targets, as measured by a remote sensor differs from the intrinsic surface radiance because of the presence of the intervening atmosphere. Since the aerosol concentration in the atmosphere varies with position and time, the amount of correction to be applied also varies. In principle, the added radiance could be removed if the concentration and optical properties of aerosol were known throughout the image. A number of methodologies have been devel­oped to provide atleast approximate corrections. (Kaufman, 1989; Slater, 1980).
The atmosphere including haze and clouds, is much more transparent to microwave than to optical and infrared region. Hence, microwave remote sensing using active sensors like Side Looking Airborne Radar(SLAR), Syn­thetic Aperture Radar (SAR) etc have all-weather capability. However emis­sion from atmosphere can affect the brightness temperature measurements of the target, even in the microwave region. It is worth mentioning that the atmospheric absorption can be advantageously used to derive atmospheric constituents and the vertical temperature profile
Interaction of Radiation with Matter
The electromagnetic radiation when incident on the earth either gets reflect­ed, absorbed, reradiated or gets transmitted through the material depending upon the nature of the object and wavelength. When the surface is smooth compared to the wavelength of incident radiation, it gets specilarly reflected in the forward direction. When the surface is rough, the incident energy is reflected uniformly in all directions which is termed as diffused. It may be noted that fine sand which appears rough in the visible region is smooth in the microwave region. Reflective/emissive properties of various surfaces at different wavelengths termed as spectral signatures are important in remote sensing since they provide information about the objects.
Concept of Signatures
Any set of observable characteristics which directly or indirectly lead to the identification of an object and/or its condition is termed signature. Spectral, spatial, temporal and  polarisation variations are four major characteristics of the targets which facilitate discrimination. Spectral variations are the chang­es in the reflectance or emittance of objects as a function of wavelength. Spatial arrangements of terrain features providing attributes such as shape, size and texture of objects which lead to the identification of objects are termed as spatial variations. Temporal variations are the changes of reflectiv­ity or emissivity with time. They can be  diurnal and/or seasonal. The varia­tion in reflectivity during the growing cycle of a crop helps distinguish crops which may have similar spectral reflectances but whose growing cycles may not be same. Polarisation variations relate to the changes in the polarisation of the radiation reflected or emitted by an object. The degree of polarisation is a characteristic of the object and hence can help in distinguishing the object. Such studies have been particularly useful in microwave region. Signatures are not, however, completely deterministic. They are statistical in nature with a certain mean value and some dispersion around it.
Spectral response of some natural earth surface features
Vegetation
The spectral reflectance of vegetation (Fig. 4) is quite distinct. Plant pigments, leaf structure and total water content are the three important factors which influence the spectrum in the visible, near IR and middle IR wavelength regions respectively. Low reflec­tance in the blue and red regions corresponds to two chlorophyll absorption bands centered at 0.45 and 0.65 mm respectively. A rela­tive lack of absorption in the green region allows normal vegetation to look green to ones eyes. In the near infrared, there is high (~45 per­cent) reflectance, transmittance of similar magnitude and absorptance of only about five percent. This is essentially controlled by the internal cellular structure of the leaves. As the leaves grow, inter cellular air spaces increases and the reflectance increases. As vegetation becomes stressed or senescent, chlorophyll absorption decreases, red reflectance increases and also there is a decrease in inter cellular air spaces, decreasing the reflectance in the near infrared. This is the reason why the ratio of the reflectance in the near infrared to red or any of the derived indices from this data are sensitive indicators of vegetation growth/vigour. In the middle infrared region of the spec­trum, the spectral response of green vegetation is dominated by strong absorption bands due to water molecules at 1.4, 1.9 and 2.7 mm.In the middle  IR reflectance peaks occur at 1.6 and 2.2 mm. It has been shown that total incident solar radiation absorbed in this region is directly proportional to the total amount of leaf water content (Tucker,1980).
Soil
Typical soil reflectance curve shows a generally increasing trend with wavelength in the visible and near infrared regions. Some of the parameters which influence soil reflectance are the moisture content, the amount of organic matter, iron oxide, relative percentages of clay, silt and sand, and the roughness of the soil surface. As the moisture content of the soil increases, the reflectance decreases and more significantly at the water absorption bands. In a thermal infrared image moist soils look darker compared to the dry soils. In view of the large differences in dielectric constant of water and soil at microwave frequencies, quantification of soil moisture becomes possible
Water
Water absorbs most of the radiation in the near infrared and middle infrared regions. This property enables easy delineation of even small water bodies. In the visible region the reflectance depends upon the reflectance that occurs from the water surface, bottom material and other suspended materials present in the water column. Turbidity in water generally leads to increase in its reflectance and the reflec­tance peak shifts towards longer wavelength. Increase in the chlo­rophyll concentration leads to greater absorption in the blue and red regions. Dissolved gases and many inorganic salts do not manifest any changes in the spectral response of water
Snow and Clouds
Snow has very high reflectance upto 0.8 mm and then decreases rapidly afterwards. In case of clouds, there is non-selective scattering and they appear uniformly bright throughout the range 0.3 to 3 mm. The cloud tops and snow generally have same temperature and hence it is not easily possible to separate these in the thermal infrared region. Hence the two atmospheric windows in the middle infrared wavelength regions 1.55 to 1.75 and 2.11 to 2.35 mm are important for snow cloud discrimination.
REMOTE SENSORS
nsors used for remote sensing can be broadly classified as those operating in the Optical-IR (OIR) region and those operating in the microwave region because the technology for developing microwave sensors is quite different from that for OIR sensors. OIR or microwave sensors can be further subdi­vided into passive or active.  Sensors which sense natural radiations, either emitted or reflected from the Earth, are called passive sensors. It is also possible to produce electromagnetic radiation of a specific wavelength or band of wavelengths and illuminate a terrain on the Earth’s surface. The interaction of this radiation with the target could be then studied by sensing the scattered radiation from the targets. Such sensors which produce their own electromagnetic radiation are called active sensors. Again, sensors (active or passive) could be either imaging, like the camera, or non-imaging, like the radiometer.
The major sensor parameters which have bearing on optimum utilisation of data include: i) Spatial resolution - the capability of the sensor to discriminate  the smallest object on the ground;  ii) Spectral resolution - the spectral bandwidth with which the imagery is taken; iii) Radiometric sensitivity - the capability to differentiate the spectral reflectance/ emittance between various targets and  iv) Dynamic range - the minimum to maximum reflectance that can be faithfully measured. In addition, the sensor should produce imagery with geometric fidelity. For engineering reasons it is not possible to simulta­neously get the best of all parameters. Hence trade-off between various parameters will be required to realize a sensor system. Let us now consider various types of sensors which are generally used for resources survey. (Joseph and Manjunath, 1983; Calla, 1983, Joseph, 1996)
OIR Sensors
Photographic cameras
Photographic cameras are the oldest and probably the most widely used imaging systems. They have been successfully used from aircraft, balloons, manned and unmanned spacecraft. Though there are different types of cameras, frame cameras have been most commonly used as remote sensors. A multiband camera enables simultaneous photography of a ground scene in more  than  one spectral  band.  This  can  be  achieved generally by using different single-band cameras and very accurately aligning them so that all the images are geometrically registered. In this case each band will have its own lens, film magazine and appropriate filters. Alternately special optics transfers images in different spectral bands on to a single large photographic film. Some of the limitations of photographic cameras are their limited spectral response (only upto ~0.9 mm) and dynamic range, non amenability to digital processing and problems associated with reproducibility of the quality of the imagery.
Television cameras
Television  cameras  were  the first imaging systems used in space to get the imagery of the earth telemetered down as  electrical signals. The basic principle of TV cameras used for imaging from space is similar to that used for commercial TV. The resolution, dynamic range of observation and the radiance accuracy essentially depend on the TV tube used. The types of TV tubes commonly used from space are direct beam read out vidicon, Return Beam Vidicon (RBV) tubes and secondary electron conduction tubes. For very low light level observation, the sensitivity of the TV camera can be increased by coupling the TV tubes to one or more image intensifier tubes
The TV cameras also have a limited spectral response depending on the material used for the photosensitive surface. With silicon diode array as target, the response can be extended upto about 1æm. Though targets with pyro-electric materials have been developed so that TV cameras can be operated in the thermal IR band (8 to 14 mm), they do not have the capability to produce high resolution imagery especially from spacecraft. TV cameras also have a limited dynamic range, usually less than 1:100.
Optical mechanical scanners
Most of the limitations seen with photograph­ic and TV imaging system are overcome in optical mechanical scanners, though they have their own limitations. Figure 6 shows the principle of operation of a line scanner. The radiation emitted (or reflected) from the scene is intercepted by a scan mirror, which diverts the radiation to a collect­ing telescope. The telescope focuses the radiation to a detector. The detec­tor receives radiation from an area on the ground (picture element or pixel), defined by the detector size and focal length of the telescope. By rotating the scan mirror which is normally inclined 450 to the optical axis, the detec­tor starts looking at the adjacent pixels on the ground. Thus, by the scan mirror rotation/oscillation, radiation is received and measured from a contin­uous line of length corresponding to the total scan angle. If such an instru­ment is mounted on a moving platform (aircraft or spacecraft) with the optical axis parallel to the platform motion, the motion of the platform produces successive scan lines, giving a contiguous imagery. In case of a multispectral scanner (MSS), the energy collected by the telescope is chan­neled to a spectral dispersing system (spectrometer) to be registered in dif­ferent spectral bands. Typical instruments using this principle include LANDSAT MSS and TM and the Very High Resolution Radiometer on board INSAT.  There is a practical limit in improving the spatial resolution using these opto-mechanical scanners. Charge Coupled Devices are currently used to provide very high resolution imagery
Linear-Imaging, Self-scanning Sensors (LISS)
In this system, the basic sensor is a linear array of solid-state detectors. The array may be made of photo-diodes, photo-transistors or charge-coupled devices (CCD). In an imaging system using LISS, the optics focuses a strip of terrain in the cross track direction on to the sensor array. The image from each detector is stored and shifted out sequentially to get a video signal like one scan line in the TV camera. The motion of the platform produces successive scan lines, thereby producing a two-dimensional picture (Figure 7). The spatial resolution primarily depends on the number of photo detectors available in a linear array and the required swath. Such sensors are expected to give a resolution of a few tens of metres even from geostationary altitudes. Currently a number of aircraft and spacecraft imaging systems are operating using CCDs. Some of the current imaging systems from spacecraft platform include French Nation­al Earth Observation Satellite (SPOT), Japanese Marine Observation Satel-lite (MOS-1) and the Indian Remote Sensing Satellites (IRS). Major specifications of sensors on board LANDSAT, SPOT and IRS are given in Table 1. IRS-P3 launched using indigenously built polar satellite launch vehicle PSLV-3 carried Modular optoelectronic sensor and three-band WiFS.  MOS comprises three cameras MOS-A operating in four narrow bands in oxygen absorption band (0.765  mm; 1.4 nm(Dl)), MOS-B operating in thirteen narrow spectral bands (Dl   = 10 nm) in visible and near infrared region and MOS-C operating at 1.6 mm. WiFS operates in red, near infrared and short wave infrared bands.Of late sensors are being developed to make spectroscopic measurement with very fine spectral resolution (<.001 mm) giving continuous coverage of the spectral region of interest. Such imaging spectrometers when operational are expected to provide additional information of vegetative cover, geology etc
OIR Active sensors
With advancement of high power laser in the optical and IR region, active laser remote sensing is promising new means of obtain­ing useful information on earth and its environment, especially related to atmospheric constituents and phenomenon. The laser system used for remote sensing is referred to as LIDAR (acronym for Light Detection And Ranging, similar to RADAR). Details of LIDAR systems are beyond the scope of the present paper.

Major Specifications Of LANDSAT, SPOT HRV 1 & 2 and  IRS LISS-I , LISS-II, LISS-III, WiFS & PAN

Satellite sensor
(year of launch)

Spectral bands
(in mm)

Ground Resolution
(m)

Quantisa-tion level

Repetition Cycle
(days)

Swath
(km)

LANDSAT 1,2,3,4,5

(1972,75,78,82,84)

MSS

0.5-0.6,  0.6-0.7,
0.7-0.8,  0.8-1.1

79

128

18
(LANDSAT 1,2,3)
16
(LANDSAT 4,5)

185

TM

0.45-0.52,0.52-0.60, 0.63-0.69,0.76-0.90, 1.55-1.75,2.08-2.35, 10.4-12.5

30

 

120

256

16

185

SPOT 1,2,3

(1986,90,93)

PAN

0.51-0.73,0.61-0.68

10

256

26

60

HRV

0.50-0.59,0.79-0.89

20

256

26

60

IRS-1A & 1B

(1988,91)

LISS-I & II

0.45-0.52,0.52-0.59, 0.62-0.68,0.77-0.86

72.5  (LISS-I)
36.25(LISS-II)

128

22

148.48
145.48
(LISS-II-A&B)

IRS-P2

(1994)

LISS-II

0.45-0.52,0.52-0.59,
0.62-0.68,0.77-0.86

37x32

128

24

131

IRS-1C/1D

(1995,97)

LISS-III

0.52-0.59,0.62-0.68
0.77-0.86,1.55-1.70

~23.5 VNIR
~69 MIR

128

24

140

WiFS

0.62-0.68,0.77-0.86

188

128

5

804

PAN

0.50-0.75

5.8

64

5(revisit)

70
Microwave Remote Sensors
Passive microwave remote sensors
Microwave radiometers are passive sensors used to measure the emitted energy. The emitted energy is collected by a suitable antenna. The signal is represented as an equivalent tempera­ture, that is,  the temperature of a black body source which would produce the same amount of signal in the bandwidth of the system. One of the popular techniques of implementing this is by using the Dicke-switched method wherein the received signal is compared with a stable reference source. The power received by the radiometer is proportional to the product of the absolute physical temperature (T) and emissivity (î).  T is referred to as brightness temperature.If the microwave radiometer is used in a scanning mode similar to the optical scanner, passive microwave imaging is possible. Scanning may be done either by a mechanical drive of the antenna or by electronically scanning a phased array.
Microwave active sensors
Side Looking Airborne Radar (SLAR) is the first active sensor used to produce imagery of the terrain from the backscattered microwave radiation. The antenna mounted sideways of an aircraft transmits a pulsed microwave energy which illuminates the ground. The return signal is received by the same antenna and is processed either on board or on ground.The antenna radiation pattern has the shape of a fan-beam  such that a narrow beam width is produced in the azimuth (along track direction), while the across direction is broad defining the swath. The radar returns scattered back from targets at different rays are separated in time at the radar receiv­er. The scattered energy depends on the radar cross section of the target, the wavelength, the slant range and the radiation pattern. Appropriate radiometric correction has to be carried out in order to produce an image, which is a true representation of the scattering cross section variation of the terrain.

The spatial resolution of a SLAR at a certain height and look angle is con­trolled by two independent system parameters, namely pulse duration for range resolution (Rr) and antenna length for azimuth resolution. Narrow pulse width demands higher system bandwidth, the engineering consideration of which limits Rr. Similarly there are  practical limitations for increasing the length of the antenna. For example at a wavelength of 5 cm, from a satellite height of 500 km to produce 100 meter azimuth resolution at a look angle of 30o an antenna of length 50 meters is required which is too big to carry on a satellite. Thus SLAR cannot produce fine resolution radar imagery from spacecraft altitudes. Synthetic Aperture Radar(SAR) overcomes this problem.

Consider an antenna carried on a platform moving at a constant velocity. The antenna beam axis is oriented at right angles to the platform velocity vector. When the antenna is at A (Figure  8) the target P is just illuminated. It continues to illuminate the target for a distance LSA until it reaches C. As the radar sends periodic pulses, the return pulse phase history during the tra­versal from A to C is stored. By a complex processing all the signals from P are added in phase. Thus the equivalent of very long antenna array with length LSA is ‘synthesised’ from a number of small elements. LSA is called the ‘synthetic aperture’ and is simply the total width of the real aperture foot print beam on the ground. The azimuth resolution (theoretical) obtained by this technique is L/2, where L is the real antenna length, while the range resolution is same as that of a real aperture radar. Thus for a SAR the resolu­tion is independent of the platform altitude and wavelength. Salient features of some of the SAR systems are given in Table 2.

Spaceborne SAR can also be operated as a scatterometer. In this configura­tion since the data rate is low, it can be transmitted through normal teleme­try channels, enabling reception by simpler ground station, to have wider coverage. The scatterometer measures the scattered signal strength, as a function of angle, frequency, polarisation or some other variables. The spaceborne scatterometer generally has coarse resolution, however, ade­quate for the intended applications such as ocean roughness, wind speed etc

PLATFORMS

The sensor systems need to be placed on suitable observation platforms. They can be stationary or mobile depending upon the needs of observation and constraints.  For an imaging system, in general, the spatial resolution becomes poorer as the platform height increases, but the area coverage increases. Thus a trade off between the resolution and synoptic view, plat­form’s ability to support the sensor system, in terms of weight, volume, power etc and the platform stability have to be considered. Though aircraft, balloons, rockets and satellites have been used as platforms, the most extensively used are aircrafts and satellites and hence our discussions will be restricted to them.

Aircrafts are mainly useful for surveys of local or limited regional interest. One of the major advantages is their ability to be available at a particular location at a specified time. They can be used at low altitudes (~ 1 km) to few tens of kilometers depending on aircraft. Currently there are aircrafts fitted with multiple sensors, capable of observations covering the whole range of the electromagnetic spectrum. The major limitation is the high cost for global / regional coverage on repetitive basis.

Earth observation from a satellite platform provides a synoptic view of a large area, which is very useful for understanding interrelationships between various features, further it can be made under known solar zenith angle providing similar illumination conditions. Another major advantage of satellite is its ability to provide repetitive observations of the same area with intervals of a few minutes to a few weeks depending on the sensor and orbit. This capability is very useful to monitor dynamic phenomena such as cloud evolu­tion, vegetation cover, snow cover etc

Major   specifications  of  some  of  the   space­borne   sar systems

SYSTEM PARAMETER

SEASAT

Shuttle  Imaging  Radar

ERS-1/2

JERS-1

RADARSAT

A

 

B

 

C

 

 

 

Launch year

1978

1981

1984

1994

1991/5

1992

1995

Altitude, km

790

250

225

225

785

568

800

Inclination
(deg)

108

38

57

57

98.5

-

98.6

Frequency (GHz)

1.28

1.28

1.28

1.25

5.3
5.3
9.6

1.275

5.3

Polarization

HH

HH

HH

Multiple

VV

-

HH

Incidance angle (deg.)

23

50

15-65

15-55

23

-

20-59*

Swath width ( km)

100

50

40

30  to  60

100

75

50 to 500

Azimuth Resolution (m)

25
(4 look)

33
(6 look)

20
(4 look)

40
(4 look)

30

18x18

25x28*

Spatial resolution, m

25

38

25

Variable

218

18

Variable
(10-100)
Sources
1Cimino J.B., and Elachi C., (1982). SIR-B Radar on the Shuttle, Proc. of Int. Symp. on Remote Sensing of    Environment,   Vol.I.

2 EOS Synthetic Aperture Radar, Vol. IIF, Page 121.

3. ERS-1 Publication, ESA BR-36, Nov. 1989
4. SIR-C Science Plan, JPL Publication 86-29, Sept. 1981
5. CEOS year book, 1995, ESA

Two types of orbits are possible (i) Geostationary and (ii) Near Earth orbit. For a satellite orbiting in the equatorial plane of earth from west to east at about 36000 km above the earth, the  period of revolution of satellite exactly coincides with that of the rotation of the earth about its own axis. Thus the satellite appears stationary with respect to the earth. Geostationary satellites are extensively used for meteorological observations. Due to the large dis­tance from earth, high resolution imaging from geostationary satellites is difficult. Resolution of about a kilometer has been successfully obtained from a number of geostationary satellites.
Near-earth orbit height varies from a few hundred  kilometers to several thousand kilometers. Most useful orbit in this category for remote sensing is the circular, near polar, sun- synchronous orbit. In a sun-synchronous orbit all points at a given latitude (say on a decending pass) will have the same local mean solar time. It must be emphasised that fixed mean solar time does not mean that the standard time will remain fixed for all points at a given latitude, because of the fact that discrete time  zones are used to determine the standard time throughout the world
Further, the ground trace of the sun synchronous satellite can be made to recur over a scene exactly at intervals of fixed number of days by maintain­ing the height of the orbit to a close tolerance, thus ensuring repetitive observations of a scene at the same local time. However, it should be noted that the solar zenith angle changes due to seasonal variations cannot be eliminated
DATA  PRODUCTS  GENERATION
The acquired data has  a number of errors due to (i) imaging characteristics of the sensor; (ii) stability and orbit characteristics of the platform; (iii) scene surface characteristics; (iv) motion of the earth; and (v) atmospheric effects. Preprocessing is carried out to correct these errors so that the inherent quali­ty of the original information of the scene (such as geometry, radiometry and information content) is retained. The outputs of the preprocessing which are available in standardised formats either in photographic or digital forms are known as the data products. Normally standard data products are generated applying geometric and radiometric corrections.
The procedures employed for geometric correction generally treat distortions in two groups i.e. those which are systematic (or predictable) and those which are essentially random. The former include the effect due to the earth rotation, the earth curvature, deviation from nominal altitude and attitude,  variations of the above deviations during the imaging of a scene, etc. Random errors arise from the uncertainty in the measurement/estimation of these parameters, and modelling inaccuracies. Uncorrected geometric distor­tions, result in relative positional distortion over the scene, and also absolute positional errors in latitude and longitude. To a first approximation it can be corrected from the measured/estimated parameters leaving the random errors uncorrected.
Radiometric  distortion arises due to nonlinearity of the detector response, responsivity variation between the detectors, pixel drop outs etc. Correction factors for sensor-related radiometric errors are normally generated by exten­sive calibration measurements during laboratory tests. When inflight calibra­tion techniques are employed, such information are also used to correct for post launch sensor degradation. When more than one detector is  used for a band, which is usually the case (6 detectors for Landsat MSS, 2048 detector elements for IRS LISS-1 CCDs), if the detector responses are not normalised by radiometric correction, one finds striping on the image
Standard data products generated from the corrected and formatted data may be computer compatible tapes (CCT) or photographic products. The data products generated contain other auxillary information required for interpretation such as longitude and latitude marks, sun elevation, date of acquisition and other relevant sensor-related information. Standard products generally have a location accuracy of a couple of kilometers. Improved geometric accuracy can be achieved by precision processing using ground control points (GCP). This method accounts for both systematic and random errors. The GCPs are permanent features identifiable on the image and their exact positions can be obtained from the standard topographical maps or by other means. Features that make good GCPs include intersection of roads, confluence of rivers, small water bodies etc.
Visual Analysis
Visual interpretation methods have been the traditional methods for extract­ing information, based on the characteristics such as tone, texture, shadow, shape, size, association, etc.  seen in a photograph. Though this approach is simple and straight forward it has some shortcomings.The range of gray values recorded on a film or print is limited; the number of colour tones recognized by the human brain is quite large still limited. The interpreter is likely to be subjective in discerning subtle differences in tones. Generation of photographic products from digital data, degrades the contrast. It is difficult to achieve precise registration of multiband and multitemporal images. It is difficult to be quantitative. Above all when large volume of data has to be analysed, it cannot meet the throughput requirements
Digital Techniques

Digital techniques facilitate quantitative analysis, use of full spectral informa­tion and avoid individual bias. Simultaneous analysis of multitemporal and multisensor data is greatly facilitated in digital methods.
In digital classification the computer analyses the spectral signature, so as to associate each pixel with a particular feature of imagery. The reflectance value measured by a sensor for the same feature say wheat field will not be identical for all wheat pixels i.e. response variation within a class is to be expected for any earth surface cover due to various reasons. Therefore the radiance value for a class will have a mean and a variance. Figure 9 shows a two dimensional plot of radiance in two wavelengths for three classes viz., for wheat, mustard and corn. This two dimensional space is called the fea­ture space. One finds a natural clustering of classes in three groups indicat­ing the signature differences of wheat, mustard and corn. When the clusters corresponding to different ground covers are distinct it is possible to asso­ciate localised regions of the feature space with specific ground covers. Such distinct clusters do not happen in real life situation. The digital classifi­cation technique essentially partitions this feature space in some fashion so that each pixel in the feature space can be uniquely associated with one of the classes. This is achieved by suitable statistical methods and a number of such algorithms are available in the literature  (Swain and Davis, 1978).

The classification techniques used can be broadly categorised as either supervised or unsupervised approaches. In the former, the analyst as a first step, called the training stage, ‘trains’ the computer by compiling an ‘inter­pretation key’. Spectral attributes for each cover type of interest are numeri­cally developed. This is generally done by examining representative sample areas of known cover type, called training areas. In the second step, called the classification stage, each pixel in the image data set is compared to each category in the numerical interpretation key. This comparison is made numer­ically, using any one of a number of different strategies to decide which category an unknown pixel belongs to. Each pixel is then labelled with the name of the category it resembles, or labelled “unclassified” if it is not simi­lar to any category. An output image data set is then generated using the category label assigned to each pixel. Thus, the multidimensional input image is used to develop a corresponding classified image of interpreted category types. In contrast to supervised procedure, unsupervised classification is based on the exploitation of the inherent tendency of different classes to form separate spectral clusters in the feature space. Unsupervised classifica­tion uses algorithms which examine a large number of unknown pixels and groups them into clusters. Each cluster is then associated with a physical category.

Geographic Information System  Techniques

GIS is a computer assisted system for the capture, storage, retrieval, analy­sis and display of spatial data.  The applications of GIS range from simple database query systems to complex analysis  and decision support systems.  Areas of applications range from natural resources management to crime control  and  near  real  time  application like fllod warning.  Analysis  models  comprise simple user defined views to complex stochastic models. Some of these are reclassification, aggregation, overlays, suitability analysis, flow models, network and route, optimisation alloccation / siting etc. Geographic Information System techniques are playing an increasing role in facilitating integration of multilayer spatial information with statistical attribute data to arrieve at alternate developmental scenarios

APPLICATION  POTENTIAL

Remotely sensed data along with ground truth information and other collat­eral data has been extensively used to survey various natural resources like agriculture, forestry, minerals, water, marine etc., and to study various physical phenomena. Since the same data base is utilised by various disci­plines remotely sensed data is ideally suited to study inter-relationship of various resources. Level at which information is available from remote sens­ing data in a particular resource area is obviously quite different. Ground resolution requirements of different applications are different. Regional geological mapping may require only coarse spatial resolution data but appli­cations in cartography or urban sprawl monitoring require very high spatial resolution data. Spectral bands required for ensuring discriminability of ob­jects for different applications can be quite different. For example crop iden­tification requires properly placed spectral bands in the red, near IR and middle IR regions. However, soil moisture determination demands use of microwave data in L or C band. Summary of spectral ranges required for different applications are given in table 3.
EVOLUTION  OF  REMOTE  SENSING
International  Scenario

Like many other technological advancements, remote sensing technology - especially aerial photography - had a quantum jump to meet the wartime needs of reconnaissance. During World War II rapid development took place for systems like False Colour IR photography, IR scanners, Radar imaging systems etc. However, systematic earth observation from space started in 1960 with the launch of Television Infrared Observation Satellite (TIROS-1), designed primarily for meteorological observation. Space photography also became available from Gemini and Apollo missions. Though the inclusion of cameras on Gemini and Apollo missions were off-shoot of the decision to land men on the moon, the US Geological Survey used these photographs to generate a general plan for repetitive surveys of the earth for resources and environmental investigations. The efforts by the survey to establish an Earth Resources Observation Satellite Programme led to the Earth Resources Technology Satellite (ERTS-1) project under NASA and the first satellite designed specifically for earth resources survey ERTS-1 was launched in July 1972. With the launch of the second satellite in January 1975, the name of the series was changed to LANDSAT.
TABLE  3 :  Applications and spectral ranges required / employed

Theme

Application

Spectral ranges
employed/ required

Agriculture,
 Forestry and
Land use /cover

Crop identification & acreage estimation

Crop condition assessment and yield estimation

Soil moisture

Drought monitoring

Land use/cover mapping

Forest fire detection

VIS, NIR, MIR, MW

VIS, NIR, MIR,TIR

 

TIR and Microwave(L&C bands)

VIS,NIR,MIR

VIS,NIR

3-4 Micrometer, TIR

Water resources

Mapping surface water  bodies

Water quality monitoring

Snow mapping

  • aerial extent
  • depth(water equivalent)

Flood mapping

VIS, NIR

Narrow spectral bands
In VIS and NIR, Thermal

VIS, NIR, MIR
Microwave

VIS, NIR

 Marine resources and
Coastal studies

Phytoplankton estimation

Fluorescence studies for Chlorophyll–a estimation

Sea surface temperature

Wetland mapping

Oil slicks

Narrow spectral bands
(~ 10nm in the VIS, NIR
at 685 nm with 5 nm resolution + NIR

TIR, Microwaves

VIS, NIR, MIR

UV, VIS, NIR and
microwaves(19.1 and 31 GHz)

Geology /
Mineral resources

Structural geology
Rock type identification

VIS, NIR and Microwaves
Narrow spectral bands in
VIS, NIR, MIR & TIR

VIS : 0.4 to 0.9 micrometer, NIR : 0.7 to 1.1 micrometer, MIR : 1.55 to 1.75 micrometre      and 2.08 to 2.3 micrometer, TIR : 8 - 14 micrometer and Microwavess : L, C and X bands.

LANDSAT 1 and 2 carried a four band Multispectral Scanner (MSS) and a three band RBV camera. In LANDSAT 3, a 5th band in thermal IR was added to MSS. Large amount of data pouring from MSS revolutionised the applica­tion of space image for various themes. An advanced scanner called Themat­ic Mapper(TM) with better spatial and spectral resolution and additional bands in middle IR was included in LANDSAT 4 and 5.

Four channel microwave radiometer carried by COSMOS-243 launched by USSR, was the beginning of spaceborne microwave remote sensing. NIMBUS - series launched by USA had a variety of microwave sensors designed primarily for meteorological applications. However, the first active microwave sensor specifically designed for ocean application was L band SAR carried onboard SEASAT in 1978. This had a capability to produce images with 25 m resolution. This was followed by Shuttle imaging radar (SIR - A & B) in 1982 and 1984 with capabilities similar to SEASAT. Subse­quently SIR-C carrying multifrequency and multipolarisation SAR, ERS-1/2, JERS,  RADARSAT carring microwave sensors have been flown

A quantum jump in the capability of space imaging was achieved with the French Remote Sensing System called SPOT. It has three band multispectral camera of 20 m resolution and a panchromatic band of 10 m resolution. There are two such cameras onboard each providing 60 km swath. One specific advantage of the SPOT system is that the view axis of the satellite is movable off nadir by ±27o thereby increasing the repetitivity of a particu­lar scene and facilitating stereo coverage. A number of countries like Japan, USSR, China operate remote sensing satellites. India joined the International community of operational remote sensing satellite operators with the launch of IRS-1A in March 1988.

One of the major experiments carried out on a global scale using remotely sensed data was LACIE - Large Area Crop Inventory Experiment jointly conducted by the U.S. Department of Agriculture (USDA), National Oceanic and Atmospheric Administration (NOAA) and NASA during 1974-78 (MacDonald, 1984), The objective of the experiment was to develop and test a method of estimating wheat production in major wheat growing areas of the world. Acreage estimation was based on the analysis of LANDSAT MSS data.Wheat yield estimates were made using agrometeorological models. Production estimation accuracy was quite good for USSR and for winter wheat in the US Great plains. However, the results were not impressive for spring wheat in USA and Canada because of strip farming and difficulties in separating confusion crops like barley, oat etc. Agriculture and Resources Inventory Surveys Through Aerospace Remote Sensing (AgRISTARS) was a follow-on programme to LACIE envisaged to address broader technical is­sues, to explore the use of data from other parts of the electromagnetic spectrum etc. Some of the specific projects undertaken related to crop condition assessment, development of yield models, soil moisture determina­tion etc. (Hall, 1982).  LANDSAT data is being operationally used for stratifi­cation of cropped area and to estimate crop acreages in the U.S. NOAA-AVHRR data is being extensively used to prepare global scale vegetation maps. (Justice et al, 1985).

There are a host of receiving stations for LANDSAT, SPOT and IRS spread over the globe, enabling remote sensing information to be extensively used by many countries for various applications. A number of mapping applica­tions for geology, land use, forest, snow cover, urban areas etc. have alrea­dy reached operational status in many of the countries. Thus remote sensing information has become vital for many countries for resource assessment and management

While the Govt-sponsored earth observation systems such as LANDSAT, SPOT, IRS, JERS/MOS have been extensively used for the survey of renew­able and non-renewable resources, commercial earth observation satellites in the private sector providing data of very high spatial resolutions in the range of 1-2 m in the panchromatic and 3-5 m in multi-spectral mode are revolu­tionising the scenario (Fritz, 1996).  These systems are distinguished by their flexible pointing ability, high geometric fidelity and very  rapid image-collec­tion to customer delivery.  Early Bird-1 of the Earth Watch Inc. was launched by the START-1, STC complex-MIHT from a Russian cosmodrome on December 24, 1997.  Unfortunately, ground controllers lost contact with the satellite. Ikonos of the space imaging EOSAT is scheduled for launch in 1998.  Major specifications of some of the commercial satellites are given in Table 4.

TABLE  4  : Broad  specifications  of  commercial  high  resolution  satellites

 

Corporation system

Earth  Watch

Orbital Sciences

Space  Imaging

Early Bird

QuickBird

OrbView-1

EOSAT

Mode

Pan

MS

Pan

MS

Pan

MS

Pan

MS

Resolution
(nadir GSD)

3m

15m

1m, 2m

4m

1m, 2m

8m

1m

4m

Spectral Bandwidths 
( mm)

.45-.80

.50-.59
.61-.68
.79-.89

.45-.90

.45-.52
.53-.59
.63-.69
.77-.90

.45-.90

.45-.52
.52-.60
.63-.69
.76-.90

.45-.90

.45-.52
.52-.60
.63-.69
.76-.90

Swath width (km)

6 x 6

30 x 30

36

36

8

8

11

11

Scene Size
 (at nadir)

36 km2

900 km2

36 x 36 km

36 x 36 km

8 x 8 km

8 x 8 km

60 x 60 km

60 x 60 km

Accuracy
       With GCPs
       Without GCPs

Horiz
6m
…

Vert
4m
…

Horiz
2m
…

Vert
3m
…

Horiz
2m
12m

Vert
3m
8m

Horiz
2m
12m

Vert
3m
8m

Stereo

In track

In track

In track

In and cross track

Earth observations from space for the past three decades have made import­ant contributions to the understanding of the planet earth and its environ­ment mostly by ‘one discipline at a time’ study. Recent research has shown the need to study the earth as a unified system of - Land, oceans, at­mosphere with its interlinkages and biogeo-chemical processes. Such com­plete understanding of the earth system requires multidisciplinary studies. The Earth Observation Systems (EOS) programme planned by NASA and envisat by the ESA envisage satellite systems carrying multiple sensors covering the entire observable electromagnetic spectrum (Earth Observing System, 1984). They are to be placed in orbiting polar platforms. The data is expected to provide valuable scientific inputs for geosphere, biosphere investigations.

Indian Scenario

Aerial photography was first used in India, in the year 1920 in a survey experiment (Bhavsar, 1980). The first use of aerial photography in an appli­cation other than land survey, was made in the year 1926, for flood assess­ment of river Indus at Dera Ismail Khan, then part of undivided India. Since then, black and white aerial photography has been widely used for map making on a scale upto 1:15,000. The aerial photographs   thus obtained, primar­ily for survey work in black and white, were also used on a limited scale for geological survey purposes and the study of river basins. However, Remote Sensing, as practiced in the present times, with the use of multi-spectral information, was first introduced in India by Professor Pisharoty and his colleagues in 1970, with the conduct of an experiment aimed at early detec­tion of coconut plantation disease (Dakshinamurty et al., 1971). Since the conduct of this first experiment in India, with the use of false colour infrared imageries, several other groups have become active in the country in using remote sensing technique for earth resources survey

The Department of Space (DOS)/Indian Space Research Organisation (ISRO) is the nodal agency for establishing an operational remote sensing system in India so as to establish a National Natural Resources Management System (NNRMS). To take care of the operational needs of remote sensing in the country, a dedicated unit - National Remote Sensing Agency (NRSA) - was established in 1975 at Hyderabad. NRSA is responsible for establishing and operating earth stations for receiving remotely sensed satellite data and associated Data Processing System. NRSA is also responsible to disseminate data to various users and to carry out application projects to meet specific user needs. NRSA currently receives data from LANDSAT, SPOT, ERS, IRS and plans to establish facility to receive SAR data from ERS-1. In addition, NRSA has dedicated aircraft fitted with various modern remote sensors and conducts aerial flights as per the user requirements.

Evolution of Indian Remote Sensing Satellite Programme

Towards realising an operational space segment ISRO is actively involved in the development of various elements such as sensors, data product, satellite platform, mission management, data reception and dissemination. The first Earth Observation Satellite of India (BHASKARA-I) was launched, in 1979. This was followed by BHASKARA-II in 1981. The BHASKARA satellites had a two-band TV payload for land applications and a Satellite Microwave Radiometer (SAMIR) for oceanographic/atmospheric applications. The spatial resolution of the image from the Bhaskara satellites was  about 1 km and  the data was used for specific applications in geology, forestry, landuse etc. The SAMIR data with “footprints”  of about 125 km has been extensively used for oceanography/meteorology.

India’s first operational remote sensing satellite IRS-1A was launched  in March 1988. IRS-1A had two types of payloads one with a resolution of 72.5 m and the other with 36.25 m, providing a swath of about 148 km. (Navalgund and Kasturirangan, 1983). The data is received at NRSA ground station at Hyderabad. Various types of user-oriented data products including standard and precision B & W/FCC photographic products as well as CCTs are generated and disseminated by NRSA. Special products, including geocoded products, are also available as per user requests.

IRS-1B identical to IRS-1A was launched on August 29, 1991. IRS-P2 carrying only LISS-II camera was launched in October 1994. The second generation satellites IRS-1C/1D have a multispectral camera with three vis­ible and near infrared bands at 23.5 m resolution and a middle IR channel of 70 m resolution. They also carry a panchromatic camera with a spatial reso­lution of 5.8 m and a wide field multispectral sensor with a coarse resolution of 188 m, providing 5 day repetitivity (Joseph, 1996).  A cartographic satel­lite IRS-P5 providing data at 2.5 resolution and along track stereo, and IRS-P6, a RESOURCESAT providing multispectral data at 6m resolution in steer­able mode are planned for 1999/2000.  IRS-P4 carrying an eight band ocean colour monitor and multi frequency microwave scanning radiometer (4 fre­quency) is planned for launch in 1999 (Table 5). In the area of microwave remote sensing, an airborne C-band SAR is under development and ISRO also plans to have a satellite system carrying various microwave remote sensors in the future. An airborne imaging spectrometer has been built and test flown

Remote Sensing Applications

Over the past two and a half decades, DOS has conducted various end-to-end application projects with various user agencies. Many of these have been taken up as national level projects encompassing large areas and vari­ous agencies. In the eighties and early nineties RS data was extensively used to prepare thematic maps on various themes / resources such as ground water, geology, flood inundation areas, forest cover, land use / cover, waste­lands, snow clad areas, watersheds, coastal, and inland wetlands, etc. Crop production forecasting using digital data is another major area of application.  Subsequent to these efforts,  the thrust is now towards inte­grating information related to various resources conjuctively with socio-economic, demographic and infrastructure data of the region to arrive at sus­tainable development plans at the local level on watershed basis. Salient outcome of some of the major projects is summarised here.

TABLE  5 :  Indian  semote  sensing  satellites  for  ocean  studies

Satellite Mission
(Year of launch or proposed)

Sensor

Spectral bands
(peak wave length in nm.)

Spatial resolution
(km)

Repetiti-vity
(days)

Swath
(km)

Quantisation
levels

IRS-P3

March 21,1996

MOS-A

756.8, 760.6, 763.5,
766.3;  1.4 nm

1.57 x 1.4

24

195

16

MOS-B

408, 443.6, 484.6,
520.8, 570.5, 615.3,
650.3, 685.3, 749.7,
868.3,1011.1, 814.1,
942.5; 10 nm    

0.52x 0.50

24

200

16

MOS-C

1600; 100         

0.52x 0.64

24

192

16

WiFS

650, 815; 60, 90

0.188x0.188

5

810

7

IRS-P4: OCEANSAT-1

(1999)

OCM

412, 443, 500, 510,       
565, 665; 20 nm, 765,
865; 40 nm

0.36x0.36

2

1420

12

MSMR

6.6, 10.65, 18.7,
21.3 GHz
Both V & H polarisation

120,75,45,40

1380

1 degree
temp.
sensitivity

OCEANSAT-2

(>2000)

Ku-Scatterometer,
Ku-Altimeter
TIR,MW radiometer

As part of the Scientific Source Finding for the Technology Mission on Drink­ing Water, hydrogeomorphological maps on 1:250,000 scale for the entire country (447 districts) have been prepared showing demarcation of potential zones/locales where there is high probability of finding ground water. Pros­pective sites for tube wells within 1.6 km radius of a village have also been identified on 1:50,000 scale maps for many of the problem villages. Feed back received from various sources indicates that scientific source finding has resulted in success rates of more than 90 per cent in drilling tube wells in hardrock areas.

Over 2800 wasteland maps, showing 13 fold wasteland categories, for 146 critically affected districts of the country (districts which have more than 15% area under wastelands) have been prepared on 1:50,000 scale at the request of the National Wasteland Development Board (NWDB). Village boundaries have been superposed on these maps to facilitate taking up development/restoration of these wastelands. About hundred additional dis­tricts have been mapped as part of Phase-II. Land use/cover maps at 1:250,000 scale for the entire country are prepared to help in agroclimatic zoning of the country. More than sixteen hundred maps showing inland wetlands have been prepared for the entire country

The project Vasundhara, covering Indian peninsula south of 17 degree lati­tude, aimed at delineating potential target areas for mineral search has led to the location of a lead-zinc occurrence in parts of Andhra Pradesh. A fairly comprehensive geographic information system package has also been devel­oped for mineral information, facilitating storage, analysis and retrieval of thematic information. The National Agricultural Drought Assessment and Monitoring System (N-ADAMS) Project provides fortnightly bulletins indicat­ing the prevalence, severity level and persistence of drought conditions taking district as a unit (Thiruvengatachari, 1988), using NOAA-AVHRR data with meteorological and other collateral information